Reminds me of
a Borgmon a bit
Born at SoundCloud
New kid in town
Different data collection
approach
I’m OK
CPU
Memory
IOpsNetwork
traffic
Disk
space
Open-box monitoring
HTTP status code
URL path
Instance
Multidimensional
sum(rate(cpu_seconds_total[1m])) by (instance)
With powerful query language
I still think it could
have been done easier...
Keeps it simple as possible
Prometheus should be as close as possible!
Reliable data collection
Shall take pairs only!
Highly available
Eh??
Kkf646a aýíz 65s s
Hěž ě 63 Wřč 6čř z6 6, 5hk.,5ř
Let me translate
it for you
Exporters
Alertmanager
2015
First public release
Federation
2016
v1.0.0
New service discovery
Community grows!
CNCF incubating
Better wait at least for
a next minor release…
Adoption!
But still, there were some
things missing…
Prometheus restarts
Highly available visualization
What data you have?
And you?
And you??
And you?!
Global data view
Do you remember anything
from yesterday?
Just last 6 hours,
after that just a blackout…
Long-term storage
Luckily, the solution was right
outside the door
2017
Generic remote read and write API added
Hello? We would like to talk with you
about your problems
Don’t forget
about us!
The uprising of remote storages
v2.0.0
New TSDB
Just catch a few bugs
Service discovery Moratorium
2018
CNCF graduation
20+ maintainers 1,000+ contributors a 13,000+ commits
Open Metrics
This standard will be
the last one for sure
2019
Subqueries
Vertical compaction
(Backfilling)
Service discovery moratorium
has been lifted
2021
Exemplars
Take a look at this exemplar
Agent mode
The Future
Sparse bucket histograms
